Eye Blinks Removal in Single-channel Eeg Using Savitzky-golay Referenced Adaptive Filtering: a Comparison with Independent Component Analysis Method

Eye blink artifact is one of the major problems in electroencephalograph (EEG) signals which mainly affected a frontal channel. A frontal channel often involved in recent applications of portable EEG devices which require a real time processing including for artifact removal. In this paper, we proposed a new referencing method in adaptive filtering for eye blinks removal of a single-channel EEG. The proposed method adopts Savitzky-Golay (SG) filter to extract the blink components from noisy EEG signals. The extracted component is then employed in adaptive filter as a reference input. We implemented adaptive neural fuzzy inference system (ANFIS) algorithm in adaptive filtering for the blink removal process. The reliability of the proposed method is demonstrated on real EEG dataset. By using the signal to noise ratio (SNR), mean squared error (MSE) and correlation coefficient as performance indicators, the proposed method is compared to independent component analysis (ICA), one of the widely accepted methods for artifact removal. The results show a low correlation between a corrected signal and a measured electrooculograph (EOG) signal, which indicates its efficiency in estimating and removing the blinks from the measured EEG signals. The results also demonstrate an improved performance compared to conventional ICA method.